Electoral Systems Matter
Decision rules matter.
To wit: the Best Picture Oscar.
Via the LAT: Oscars 2015: Why the film with the most first place votes might not win best picture.
This discusses the uses of the preferential vote (also known as the alternative vote and the instant run-off) to select the Best Picture winner since the Academy expanded the number of potential nominees in the category to up to ten six years ago.
The LAT headline is correct in one sense, but incorrect in another.
It is correct insofar as the winner will not be decided by plurality (i.e., a simple case of the movie with the most votes winning and being done with it at that). However, if one thinks about it, that would not be a desirable process. There are currently eight candidates for Best Picture and if support for those movies was relatively evenly distributed, a winner could be declared with roughly 13% of the vote. That would hardly be indicative of the which film the voters, collectively, thought was best.
If the goal is to determine the overall preference of a group of voters, then straightforward plurality voting is problematic when there are more than two candidates as it can mean that the winner could be opposed by a majority of voters (and this potential distortion gets higher the more candidates that are involved). So, for example, if there are eight movies and one movie gets 13% of the vote and the remaining 7 movies each get 12.4% of the vote, then the “best” picture would be one for which 87% of voters did not vote.
The incorrect nature of the headline is the implication that, somehow, the number one preference of the voters would not win but, rather, that a “complicated” and “counterintuitive” system (to use descriptor from the video and the written portion of the piece) is being used. However, as noted above, what this system does (as it does in places that it is used to elect candidates to office, such as in the Australian House of Representatives) is allow voters to rank their preferences in such a way that one would vote if there was a two-step run-off in which voters only chose among the top-two voter-getters (and, indeed, this system is better at capturing true preferences since one’s second place preference might not be one of the top two–a fact that could very much matter in a close, multi-candidate race).
Clearly, the elections nerd in me finds all of this interesting just because of the topic. However, I would go a step further and note that it is illustrative of the general ignorance we have in the US over various way to conduct elections that we seem able to only conceive a system which has plurality winners (even though AV is used in some local elections). Regardless, it is important to understand that there are better ways of finding out the true preferences of a group of voters than simply “he who has the most votes wins.” And before anyone thinks I am arguing for a system that prefers minority views over that of the majority–go back to the eight candidate, evenly distributed example above.
Put another way: if twelve friends are all deciding what to do for dinner and three want pizza and the other nine want various other things, what is the most democratic outcome: letting three decide for the twelve, or further deliberating and figuring out that when second preferences are taken into account that nine could go for Chinese?
This goes back a ways, but didn’t Loni Guinier get excoriated for suggesting something similar to the Oscar type voting system for the US back in the Clinton Administration? She lost a deputy AG nomination over it?
I’d appreciate your take on what that was all about, Professor, or a pointer to a good discussion on the issue. Thanks in advance.
@stonetools: As I recall, the controversy had several elements, including advocacy of a voting system (which, I think, was AV). She also made arguments about representation in terms of minorities representing minorities (i.e., sectoral representation).
Without a doubt her recommendations for electoral reform were treated as very strange (even though, if I recall properly, they were not especially exotic).
My recollection is that Lani Guinier got in trouble (in part) for advocating cumulative voting in at-large elections.
Without reading/viewing the linked item, Dr Taylor’s post seems to me to be pretty much identical to the ‘jungle primary’ that has recently been instituted in CA. So will hope to read some comments from the left-coast citizens who post here.
In general I rather mistrust this kind of decision process but it does seem like the only alternative to the parliamentary system in (a current example) Israel in which numerous small parties are necessary to form what we would call an administration (and the rest of the world calls a ‘government’). So in a multiply fractured voting public perhaps it’s the best system.
We lost a lot in the US when our two parties became divided along ideological lines. When both parties had a significant range of conservatives and liberals a dominant party still was required to do compromises and balancing acts in order to pass bills and administer government. There was a slowing of response times I guess but probably it made for a superior national system. Or maybe I’m just growing old and remembering things with a benevolent glow that they didn’t have in reality; I was certainly an ideological character who chafed at the liberals (when I was a conservative Repub.)
@JohnMcC:
Just as a point of clarification: the “primary” system in California (scare quotes, because it really isn’t a primary) is called the “Top Two” and is not what an example of AV. The Top Two system simply narrows one group of candidates down to two (it is a type of run-off system). It is not a typical runoff system insofar as in such a system if one can get 50%+1 in the first round, there is no second round.
Israel’s system is one of a large multi-member district (only one, in fact) in which seats are allocated by the D’Hondt highest averages system (a proportional representation system).
Except that, as you mention, we do have runoffs. So we’re familiar with other concepts. I think the best way to sell IRV is as an extension of the concept we are familiar with.
Ideally, I’d love to see preferential voting. I’d be more excited about it if it weren’t for the “voter intent” swamps. If we cant reasonably expect voters to punch a chad or fill in a bubble, I’m not sure what kind of preferential ballot we can trust them with.
But I think the main source of opposition is likely to come from the two major parties who benefit from telling people that if they vote third party they are throwing their vote away.
@Trumwill:
Oh, one can find a number of variations in local elections. However, on balance, the general level of understanding of election beyond “he who gets the most votes wins” is pretty low.
(It isn’t as if, for example, there is much in the way of public debate on these topic. The Guinier incident above helps illustrate this in fact. As does press coverage of voting (whether for Best Picture or of elections around the world).
I do not see why Americans would think such a system strange, many sorts awards like the Heisman Trophy use a similar system where second and third place votes matter. There is also the crazy system for voting someone into the Baseball Hall of Fame.
I have always found these types of popularity strange because how many of the voters have actually bothered to watch all eight (or even ten pictures). At least in some of the other categories, the voters are associated with that category.
Dang it Steven – I’m heading out for Chinese food, back in a bit.
@Steven L. Taylor: Runoffs aren’t exactly a rare anomaly used by some local elections here and there. The governorship of California uses them! They are also used heavily in primaries. They’re also used for federal positions in some states, just not the presidency. I’ve found that people are often more surprised when there isn’t a runoff than when there is.
I’d argue (both because I think it’s true and because it helps explain it) that IRV is still about “Whoever gets the most votes wins” it’s just a different form of inputs, similar to a runoff but better, less expensive, and less hassle.
@Trumwill: You are kind of making my point: you are correct, run-offs are not that rare, but IRV is considered too exotic to really talk about (although it, too, is used in some local elections).
Indeed, IRV is just a small modification of single seat plurality winners–so, the fact that it is difficult to get a serious conversation rather makes my underlying point.
@suerdestroyer: It is a puzzle in some years as to how in the world some movies got best picture. Some of the “best picture” winners did not show up at many theaters. Here are some movies that should have been best picture: “Raiders of the Lost Ark” , “Star Wars”, “Avatar”, “Deliverance”, and “Dr. Zhivago” (lost out to “Sound of Music”, two great movies).
@Tyrell:
I read the theory once that most of the voters do not actually see movies in a theater but watch them at home first using VHS and then using disk. A big effects laden movie does not look as good on an old 25 inch TV versus some small story movie. And as many people have pointed out, movies about Hollywood or the Holocaust usually do better in the voting.
@Steven L. Taylor: The part I was objecting to was the “why” as much as anything. Why we haven’t considered the IRV. I don’t think it’s because we’re conditioned to accept plurality victories. If we were, we wouldn’t have runoffs. But we do.
The tricky part about IRV is while it doesn’t change how votes are calculated, it does change how they are cast. We have comparatively low tolerance for tossing ballots and feel bad for voters who disenfranchise themselves out of confusion. We don’t want gramps and gramms confused in the voting booth.
The good news is that we have a federal system. So we don’t have to win over the entire country. We can start with Maine, which needs to do something about its plurality elections anyway.
That does bring us to the second problem: Such reform is not in the interest of the established players. Maybe advocates need to find a state with referenda or something. If you can overcome the first problem, I don’t actually think it’s a terribly difficult sell to make. Or maybe it is, for a country that still uses the penny. 🙂
@Trumwill: In truth, I do not think that IRV at, say, the Congressional level would make much of a change to the system (for any number of reasons). Indeed, the non-competitive nature of most House districts, for example, would mean that in most districts there would be no need to consult second preferences since the winner would win on first preferences.
IRV, however, would makes sense from a fiscal and representativeness POV if it were used in states that had run-of primaries (much of the South) or the rare case that requires an absolute majority to win office (e.g., Georgia for Congressional races).
@suerdestroyer: It’s extremely unlikely that the Academy voters at the time of Dr. Zhivago, Deliverance, Star Wars or even Raiders were watching screeners rather than seeing the films in theaters…
@Steven L. Taylor: Thank you, sir. Obviously there’s yet another item to add to my self-syllabus.
@wr:
Raiders of the Lost Ark was about the time that I was working in a video store, so yes, it was likely that the voters started watchinig at that time. Before that, it could have been home theaters (for real) or special screenings.
@Steven L. Taylor:
While most certainly true at the outset, I’d think IRV would have an impact over time, as third party candidates gained viability. Three dimensional gerrymandering has got to be harder to pull off.
As long as we have the presidency and the electoral college, we’re going to effectively have a two party system. Where I could see IRV having an impact, if it becomes widespread enough to pass the tipping point, is actually in gerrymandered districts and non-competitive states. Potentially creating elections between the Republican Party and Tea/Constitution Party in some districts, and Democratic vs Green/Socialist Party in others. Or alternately, Liebermanesque “Independent Democrat/Republican” vs Democrat/Republican.
Primary losers will have less incentive to accept defeat. One of the biggest justified arguments against them (that they will throw an election) will be lost.
The end result wouldn’t be the changing of congressional hands (Lieberman caucused with the Democrats, Constitution Party members would caucus with Republicans), but it would still be different.
I’m not saying that’s what would happen, but it could. It could also undermine the primary system entirely. Whatever changes occur, though, would be to the detriment of the two parties. I’d think, anyway. That’s why I think the parties would oppose it. For the same reason they tend not to be fans of blanket primaries.
@suerdestroyer: “Raiders of the Lost Ark was about the time that I was working in a video store, so yes, it was likely that the voters started watchinig at that time. Before that, it could have been home theaters (for real) or special screenings.”
Shockingly, the Academy in the 80s did not say “Hey, some racist creep works in a video store now, so we’d better start sending screener tapes out to our members.” In fact, the very notion was controversial with the Academy and the studios for decades.
@wr:
However, the idea that the voters were not watching movies on VHS when Raiders of the Lost Ark came out is not supported by the historical data. Raiders was released in June 1981 and by that time the world had settled on VHS tapes. And if anyone was going to have access to a video tape player, it would be the people in the movie industry. If nothing else, many of the voters probably had to go to special screening to see all of the nominated films and that would be a different experience than watching an action movie in a full theater.
The Awards are over and no big surprises. The line up this time was one of the weakest ever.
@Trumwill: I think if you look globally you will find that the main driver of the number of parties is the system used to elect legislatures, not the system to elect executives. This is not to say that there is no relationship, but without a doubt, the main reason we have a two party system in the US is the use of a single seat plurality system to elect members of the legislature as reinforced by primaries.
IRV would not change that dynamic. Even under those rules, the incentive would be to win the nomination of a major party rather than trying to go the third party route.
@Steven L. Taylor: My view is that we will effectively have a two-party system as long as the presidency is determined by the electoral college (and the procedure for when no one gets a majority of electoral votes). Everything else will flow down from that forever and ever amen (until or unless that is changed). Even if we had a proportional representation system, it would still mostly come down to two coalitions, if not only two parties on the ballot.
What I am talking about here is still within the two-party dynamic. It’s just that, instead of having a few independent congresscritters, we’d likely start to get more. Mostly congresscritters that are solidly in one camp or the other when it comes time to caucus. This could be argued to be a distinction without a difference, and if one’s goal is a truly multiparty system that’s a pretty fair way of looking at. But it’s not quite the way I look at it.
In my view, the dynamics of primaries and elections change if Chris McDaniel or Mike Castle have a more reasonable shot at garnering support and winning after losing in the primaries. Opponents won’t be able to talk about splitting the vote. There will be less fear around supporting their candidacies. More chances will be taken, and more victories will occur. The party label, while still important, may nonetheless become less so.
(In my own sake, it’s a part of a series of reforms I’d like to see to create more party fluidity. I’m fine with a two-party system. What I think is troublesome is that the parties are hard-coded in. While in Britain the Liberal Party was replaced by Labour and in Canada the Liberals look poised to be replaced by the NDP, we’re still with the Republicans and Democrats. All of which will sort out eventually – the parties morph to fit new electorates – I think it would be better if the specific parties themselves had more to worry about when and if they become uncompetitive.
I doubt this alone would be able to fix that. But it would be a step in that direction. Next up: fusion tickets! I know, it’s a pipe dream. And it’s entirely possible that I am overly optimistic on the degree to which this would make that more feasible. But Duverger’s Law has its limitations, at least in short runs.)
@Tyrell:
It’s not a puzzle, really. The Oscars are not the box office awards, nor are they the people’s choice awards. They are nothing more than “this is the movie that the voting members of the AMPAS liked best”. It’s akin to the members of a country club voting for man of the year.
The mistake that many make is in presuming that those members care, or even should care, what the public wants. It’s very much a private party.
(Sidenote: is that LAT piece, which I did not read, yet another extended whine about how American Sniper somehow got snubbed?)
@HarvardLaw92: Top five: “Lawrence of Arabia” (one of the best ever), “Patton” (great acting from all), “Sound of Music”, “French Connection” (bar scene is great), “Lord of the Rings”. Biggest misses: “Great Escape” and “Deliverance” (Burt Reynold’s best)
. Biggest mistake: “Midnight Cowboy”, a real flop.
Some years they just get it wrong, and I would say most of last several years. It is bad when you can’t find the best picture winner showing at any local theater. (“Slumdog Millionaire”, “Crash”- never even showed up on tv).Many of the best picture winners of the last few years have been box office flops. Evidently the people who make the selections are out of touch. I get it that they are not making selections based on popularity, but some of their picks have just been plain awful, so bad many of the theaters would not show them.
@Trumwill:
Yes, but there is some comparative evidence that counters this proposition: Latin America systems that have presidential systems yet have PR for legislative elections and what we find are multi-party systems.
Indeed, we find that the variable most significant world-wide to determining the structure of the party system is the process to elect the legislature. This is not to say that other factors are unimportant, of course.
@Trumwill:
I am not sure what you mean by this. My initial reaction is that we have rather fluid parties (see:, e.g., Jeffords, Jim and Specter, Arlen, amongst others–or, for that matter, Paul, Ron).
@Tyrell:
The point is, however, how can a a poll of the opinions of a certain group of people “get it wrong”? That is: all this is is the summation of the subjective views of X numbers of voters. How can an opinion be “wrong” save in the since that one does not agree with it?
@Trumwill:
Let me ask this: what is the better (i.e., easier) route for a Tea Party type in a heavily Republican district: win the primary (which has low turnout and often is dominated by more ideological voters) or run as an actual party to try and defeat the Republican by being able to get 50%+1 of the vote?
The answer is: win the primary and then coast to victory in the general.
Primaries short-circuit the impulse for party factions to run candidates in the general. It is far easier to win an internal nomination fight and then coast to victory than it is to win in the general (it is one of the reason Ron Paul, a Libertarian at times only won office as a Republican).
The primary system is a key factor in explaining why the US has the two party system it has and IRV would not incentivize factions to become parties.
The only way to incentivize factions to become parties is to introduce some type of proportional representation into the system.
@Steven L. Taylor: It’s not just having a presidency. It’s having a presidency determined by an electoral college. And not just that, but an electoral college that, in the event of a non-majority, ends up in the House of Representatives with states voting on an equal basis. Without significant other reform, I don’t see anything other than what is effectively a two-party system.
By “fluidity” I mean I mean that parties are replaced by other parties. The Liberal Party is replaced by the Labour Party. The PC Party is replaced by Conservative Party and possibly the Liberal Party replaced by the NDP. IRV won’t get us there, but it’s a step in the right direction. And steps are built on steps.
The easier path to congress in your scenario will be as a Republican. Which is why most Republicans will be Republicans. But IRV will grease the wheels on another route if for some reason you can’t win the party’s nomination. I would expect it to start with primary losers, like Lieberman or Murkowsky, except with more frequency. Then I expect people like Orman to become more common. Still mostly Republicans and Democrats, but the odds of getting elected as something else will change, and I think there’s a really good chance that the congressional totals will reflect that.
@Trumwill:
As long as you have primaries of the type that we have, the incentive will be for factions to take over existing parties rather than form new ones.
@Trumwill:
I think that you are both dismissing the comparative evidence and (more importantly) not seeing the way in which party systems interact with legislative elections.
I understand the point that you are making about the potential role of the House in the EC, although the main issue for the EC is that it has plurality winners in multi-seat districts. This leads to an incentive for two party competition far more than does the threat of the House making a decision.
So yes, the EC drives presidential competition to a binary proposition. This would influence legislative parties even in the context of PR. But, it would not be a sufficient force, in and of itself, to bind the legislative parties to a two party system if there was significant proportionality in the system.
All systems, I would note, value a majority at some point (e.g., picking a PM and cabinet) but that fact does not translate into two party systems globally.
@Trumwill:
What would suggest that IRV would grease said wheels?
Only if one can lose the nomination and then break off to run.
And, even so, that is not third party formation.
BTW: the closest thing we have, off the top of my head, to a natural experiment along the lines of what you are suggesting is the Two Top system in California. It not producing the type of results you are describing.
(It is not an ideal experiment, as it does not also have a nominating mechanism beyond essentially self-nomination).
BTW: there was a reason that the Conservatives were willing to let the LDP have their vote on AV (i.e., IRV) because, really, it would not have hurt the two large parties all that much.
The House of Representatives’ role in the process is the last firewall. But it’s the reason that Ross Perot was never, under any circumstances, become president. It takes the likelihood of a third-party presidency from the highly-improbable to a virtual impossibility. It would take considerably more than multimember districts to overcome that. And as long as we have two candidates for president, we’re going to have two major parties (or coalitions, a Liberal/National type situation is possible). That’s why I mention it. Even if you can somehow manage to get past everything else (something I would agree is very, very difficult), I do not see how you can get past that.
With that wall, it’s hard for me to look at our system and say single-member congressional districts are the thing keeping us at two parties, and preventing us from having an LDP or Alliance or an NDP even for a while. It’s the ultimate trump card.
It greases the wheels by removing a barrier. Even with all of the barriers in place, with the pits and perils of 3-way elections and “wasting your vote”… independent candidates can and do get elected. Take away the “wasting your vote” argument, and I believe there is a good chance that more do.
I never really thought this the case for California (or Louisiana) for a couple of reasons. First, because it’s not a sufficient change (IRV is less than MMD, but blanket primaries are less than IRV). I do think that if you combine blanket primaries with IRV, some interesting things might happen (and I think IRV could be the undoing of primaries as we know them). Or might not, because: Second, while California is our most populous state, it’s only one state. I don’t think any single state with the IRV will necessarily show any deviation from the norm on account of it. Because it’s still the norm. That’s why I initially spoke of tipping points.
@Trumwill: Australia used IRV to elect its first chamber and has an effective number of parliamentary parties of 2.28 (the average from 1990-2010).
During the same period the US has an ENPP of 1.98 and the UK 2.32.
That does not provide much evidence for your greasing hypothesis (quite the opposite).
@Tyrell:
Hence the People’s Choice Awards. The people voting for the Academy Awards don’t care what you think.
@Steven L. Taylor: Does that count the Coalition as one party, two parties, or four parties?
Either way, I believe that Australia actually supports my belief that it would encourage or at least enable alternative parties. The continued existence of the National Party*. There is only one state (and one territory) where the two have merged, and that happens to be one of two states that do not use IRV for state ballots*, and it was costing them elections.
Different countries are different. The same country changes over time. The landscape in Canada and the UK shouldn’t look like they do, but they do. Our two-party system isn’t going anywhere, but holding everything else constant a system in which you can vote your heart before voting your mind seems more likely to me to be suppor tive of independent candidates than one where that’s “wasting your vote” or “throwing an election”.
* -There is the argument that National/Liberal are functionally the same party even if they have different names, but (where both exist independently), but they field different candidates and report to different parties. It’s actually not that far from what I would like to see. It’s actually close to describing what I would consider an ideal result.
** – They split the difference, with voters having the option but bot the requirement to vote for more than a single candidate. Arguably, this is the system we would go with – grampa’s vote, again – which might inhibit splinters within coalitions.
@Trumwill: If you count the Liberals and the National Party as two parties the ENPP’s is 2.516. If you treat them as the same party the result is 2.045. We took the mean of the two to generate the 2.28 number.
But, really, if I understand your position, you would be happy with a situation like that in UK. (Which, I would note, is one of single seat district, plurality winners and no IRV). Along those line, I would further note that the major difference between the US and the UK, electoral system-wise, is that the US uses primaries to nominate candidates and the UK does not (see my points above).
@Steven L. Taylor: Not quite, the UK is something I’d really like to avoid. That the two center-left parties get a majority of the vote but the center-right party gets power because they got the plurality? Ugh. Assuming, as I do, that most of the LDP would prefer Labour over Conservative.
My own preferences tend to run a cross-section of what multiple other countries do. I like a multiparty system with announced coalitions (like New Zealand) but prefer a two-party system over a multiparty system where coalitions are determined only after the results are in. That strikes me as more of a “custom” thing than a “law” thing, though.
I’d support multimember districts in the US at the state level (we have two houses, let’s have one of them be MMD instead of just bigger districts), but I’m skeptical of it at the national level due mostly to population making it unfeasible (I like districts, preferably with fewer than a million people in them). If we didn’t mind a House with membership in the thousands, I’d be down with mixed districts or maybe even just proportional.
I really wish that we had a degree of separation between state parties and national parties, like Canada (Saskatchewan, for example) and Australia. That’s not something that can be legislated, of course, though having parties within coalitions could help facilitate that maybe.
Out of curiosity, for Australia, when counting them as two parties, how did you count the LNP and CLP? The former as Liberal and the latter as National, or by the membership of those actually seated? (I’m curious about SOP, not questioning your numbers.)
To quote Kramer from Seinfeld, this post is scratching me right where I itch. I have long been in favor of us moving from first-past-the-post elections to something along the the lines of instant runoff voting or Condorcet.
I live in Louisiana now and we have “jungle primaries”. In practice, it has meant that we had to wait to see Mary Landrieu lose a month after the original election. I would have much preferred using IRV and knowing sooner.
@Rob Prather: Regardless of anything else, it makes all the sense in the world to use IRV in any run-off situation since it saves money on a second round and, more importantly means the the “first” and “second” rounds consist of the same voters, which is never the case in a straightforward run-off.
@Trumwill:
We counted the LP, NP and CLP together as the Liberal/National Coalition for the calculation that counted them as one party.
@Steven L. Taylor: Steven, as is often the case, we are in complete agreement. I’m a good example. I didn’t vote in the runoff because I knew Landrieu would lose. I got up and voted at 8:00 on a Saturday for the actual election, but couldn’t be bothered a month later. Far more efficient to do it on one day and have the same voters make the decision.
@Steven L. Taylor: On a side note, I would have loved to have a chance to meet you when you were in NOLA a few weeks back. Neither Chris nor I thought about it until the last second.
@Rob Prather: That was a shame. Chris texted me literally within five minutes of my wife and kids showing up (they came down for the weekend).
Next time I will be sure to get together with you.
@Steven L. Taylor: It will indeed be my pleasure. I remember him texting you and thinking “if I’d known they were at a political science conference I’d have asked about Steven?”